Cancer is the second leading cause of fatalities among U.S. children under the age of fifteen. Exposure to pesticides is potentially an important cause of childhood cancer. Contaminated drinking water is suspected as a source of such exposure. However, there is no direct evidence that implicates pesticide-contaminated water as a risk factor. The objective of this application is to estimate the risk of specific childhood cancers with each watershed in Texas and the role that 'local' cropping practices have in contaminating water supplies. The central hypothesis is that childhood cancers have incidence patterns that correspond to the agricultural use of pesticides and that a proportion of the risk is attributable to watershed-mediated exposure to agricultural pesticides. This hypothesis will be tested by three specific aims, which are strongly supported by preliminary analysis of the Texas Cancer Database. 1) Estimate the cancer risks attributable to cropping practices. The product of this aim will be a GIS layer of cancer risk based on county-level crop production. 2) Estimate the risk for specific childhood cancers in each watershed in Texas. The GIS layer of cancer risk produced here will be based on risk estimates for discrete watersheds. 3) Estimate the extent that the risk surfaces for watersheds and cropping practice are correlated. This aim will utilize formal Bayesian analysis of homology of the two GIS layers. The approach is innovative, because it capitalizes on recently developed Bayesian mapping and analytical techniques. The proposed research is significant, because we expect to estimate the potential for contaminated watersheds to mediate childhood exposure to pesticide agricultural carcinogens. Data acquired will be useful in further study design to select sampling sites that will optimize environmental and biomarker testing. It is anticipated that the results produced will provide the preliminary data needed to continue definitive investigations under the auspices of subsequent R01 funding.